YouGov :"POLL - Blunkett's proposals on internet privacy: for or against"

[Richard D G Cox]

On Fri, 14 Dec 2001 13:14:45 "Caspar Bowden" <cb@fipr.org> wrote:
> ... seems arguable that newsgroup (not article number) might fall
> within defn. of traffic data - but I wouldn't care to put money on it.

If I understand the newsgroup mechanism correctly, since specific articles
can, and far-too-frequently do, exist in more than one newsgroup: initial
requests to the NNTP server are for a list of article numbers that contain
the specified newsgroup header and received-dates within a specified window
(that's akin to a keyword search) and the subsequent request is for the
headers (and/or bodies) of articles that are specified (usually by number)
in that request.  *Those* last requests do not refer to actual Newsgroups.

(I would not have expected the *results* of a keyword search to fall
within the definition of traffic data ... the fact that my client has
asked for a list of article numbers is not the same as downloading them)
#
If my client asks a server for articles in, say, cam.misc between last
Monday and today, I *may* get back the number of an article, say 7151992,
which for some reason exists both in cam.misc and in some other unrelated
newsgroup - say alt.amsterdam.porn.  If I want to read that article, my
client will later ask that NNTP server for article 7151992, and as NNTP
servers are stateless it will NOT know from which newsgroup I was trying
to download that article.  If - to meet LEA requirements - it has to look
up and store the name of the newsgroup from the header, and so reports
that I have downloaded it from alt.amsterdam.porn, that information would
be completely inaccurate - and also possibly defamatory.

If, of course, I have misunderstood the mechanism here, I guess my inbox
is going to get fuller than usual. ;-))

-- 
Richard D G Cox <Richard.Cox@mandarin.com>